How would I avoid using two separate queries for this purpose? - mysql

Question

I am using MySQL to create a database of articles and categories. Each article has a category. I would like to make a feature for the admin panel that lists all the categories, but also includes the latest article for each category. The method I usually use is to fetch rows from the category table, loop through the results, and then create another query using something like FROM articlesWHERE category_id = {CATEGORY_ID} ORDER BY article_id DESC LIMIT 1. That method just seems like overkill to me and I am wondering if it can be done in one query(Maybe with joins and subqueries?).
This is the current query I have that fetches categories:
SELECT * FROM article_categories ORDER BY category_title ASC
These are my tables:
CREATE TABLE IF NOT EXISTS `articles` (
`article_id` int(15) NOT NULL AUTO_INCREMENT,
`author_id` int(15) NOT NULL,
`category_id` int(15) NOT NULL,
`modification_id` int(15) NOT NULL,
`title` varchar(125) NOT NULL,
`content` text NOT NULL,
`type` tinyint(1) NOT NULL,
`date_posted` timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,
`status` tinyint(1) NOT NULL,
`attachment_id` int(15) NOT NULL,
`youtube_id` varchar(32) DEFAULT NULL,
`refs` text NOT NULL,
`platforms` varchar(6) NOT NULL,
PRIMARY KEY (`article_id`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1;
CREATE TABLE IF NOT EXISTS `article_categories` (
`category_id` int(15) NOT NULL AUTO_INCREMENT,
`parent_id` int(15) NOT NULL,
`title` varchar(50) NOT NULL,
`description` text NOT NULL,
`attachment_id` text NOT NULL,
`enable_comments` tinyint(1) NOT NULL,
`enable_ratings` tinyint(1) NOT NULL,
`guest_reading` tinyint(1) NOT NULL,
`platform_assoc` tinyint(1) NOT NULL,
`allowed_types` varchar(6) NOT NULL,
PRIMARY KEY (`category_id`,`title`)
) ENGINE=InnoDB DEFAULT CHARSET=latin1;
This is the query I have come up with so far:
SELECT
c.category_id, c.title, c.description,
a.article_id, a.category_id, a.title, COUNT(a.article_id) AS total_articles
FROM article_categories AS c
LEFT JOIN articles AS l ON (
SELECT
article_id AS article_id, category_id, title AS article_title
FROM articles AS l
WHERE l.category_id = c.category_id
ORDER BY l.article_id
DESC LIMIT 1)
LEFT JOIN articles AS a ON (c.category_id = a.category_id)
GROUP BY c.category_id
ORDER BY c.title ASC
The above query gives me the following SQL error:
Operand should contain 1 column(s)
Why is this happening?

Answer 1

You can return list of all the categories and recent article in each category using one query, Try this
SELECT C.*, A.*
FROM article_categories C
LEFT OUTER JOIN articles A ON c.category_id = A.category_id
WHERE
(
A.category_id IS NULL OR
A.article_id = (SELECT MAX(X.article_id)
FROM articles X WHERE X.category_id = C.category_id)
)

Answer 2

This will restrict the articles to just the highest article_id per category and make use of the indexes on those tables:
select
ac.category_id, ac.title, newa.article_id, newa.title article_title
from article_categories ac
left join articles newa on ac.category_id = newa.category_id
left join articles olda on newa.category_id = olda.category_id
and olda.article_id > newa.article_id
where olda.article_id is null
;
See this Demonstrated at SQLFiddle

Answer 3

Shoelace, I was browsing your other questions and saw that this was unresolved so I've decided to take a crack at it.
This is a little tricky, but I don't think it's too bad, assuming I understand your question correctly. First, get the latest article date for each category:
SELECT a.category_id, MAX(a.date_posted)
FROM articles a
JOIN article_categories c ON c.category_id = a.category_id
GROUP BY a.category_id;
Then, join that with your articles table on the condition that the category_id and date are equal and you have what you need:
SELECT ar.*
FROM articles ar
JOIN(SELECT a.category_id, MAX(a.date_posted) AS latestDateForCategory
FROM articles a
JOIN article_categories c ON c.category_id = a.category_id
GROUP BY a.category_id) t
ON t.category_id = ar.category_id AND t.latestDateForCategory = ar.date_posted;
SQL Fiddle.